Child's Nervous System (2020) 36:73–86

https://doi.org/10.1007/s00381-019-04263-4

FOCUS SESSION

Cerebrospinal fluid dynamics in pediatric pseudotumor

cerebri syndrome
Afroditi-Despina Lalou 1,2 3 1 2
& James S. McTaggart & Zofia H. Czosnyka & Matthew R Garnett & Deepa Krishnakumar &
3

Marek Czosnyka 1

Received: 22 May 2019 / Accepted: 17 June 2019 / Published online: 19 July 2019
# The Author(s) 2019

Abstract
Purpose There is a growing body of evidence highlighting the importance of comprehensive intracranial pressure (ICP) values in
pseudotumor cerebri syndrome (PTCS). Due to the highly dynamic nature of ICP, several methods of ICP monitoring have been
established, including the CSF infusion study. We have performed a retrospective review of the CSF dynamics measurements for all
pediatric patients investigated for PTCS in our center and examined their diagnostic value compared with clinical classification.
Methods We retrospectively recruited 31 patients under 16 years of age investigated for PTCS by CSF infusion test. We used the
clinically provided Friedman classification 13/31 patients with definite PTCS (group A), 13/31 with probable PTCS (group B),
and 5/31 not PTCS (group C), to compare CSF dynamics in the 3 groups.
Results CSF pressure (CSFp) was significantly increased in group A (29.18 ± 7.72 mmHg) compared with B (15.31 ±
3.47 mmHg; p = 1.644e-05) and C (17.51 ± 5.87; p = 0.01368). The amplitude (AMP) was higher in the definite (2.18 ±
2.06 mmHg) than in group B (0.68 ± 0.37; p = 0.01382). There was no in either CSFp or AMP between groups B and C. No
lower breakpoint of the AMP-P line was observed in group A but was present in 2/13 and 2/5 patients in groups B and C. In group
A, sagittal sinus pressure (SSp) and elasticity were the only parameters above threshold (p = 4.2e-06 and p = 0.001953, respec-
tively), In group B, only the elasticity was significantly higher than the threshold (p = 004257). Group C did not have any of the
parameters raised. The AUC of CSFp, elasticity, and SSp for the 3 groups was 93.8% (84.8–100% CI).
Conclusions Monitoring of CSFp and its dynamics, besides providing a more precise methodology for measuring CSFp, could
yield information on the dynamic parameters of CSFp that cannot be derived from CSFp as a number, accurately differentiating
between the clinically and radiologically derived entities of PTCS.

Keywords Idiopathic intracranial hypertension . Pseudotumor cerebri . Intracranial pressure . CSF dynamics . CSF infusion
studies

Introduction raised BMI and its accurate diagnosis is particularly challeng-

The pseudotumor cerebri syndrome (PTCS) has been princi-
pally described as a disease of post-pubertal females with
* Afroditi-Despina Lalou
adl43@cam.ac.uk
1
Department of Clinical Neurosciences, Division of Neurosurgery,
University of Cambridge, Hills Road, Cambridge CB20QQ, UK
2
Department of Clinical Neurosciences, Division of Neurosurgery,
Cambridge University Hospital, Cambridge, UK
3
Department of Pediatric Neurology, Cambridge University Hospital,
Cambridge, UK
ing in the pediatric population. The current intracranial pres-
sure (ICP) thresholds in the pediatric spectrum of the disease
(28 cm CSF for obese and/or sedated children, 25 cm CSF for
normal weight and non-sedated children for definite PTCS)
[1–3] have been established epidemiologically, being set at the
90th percentile level of ICP recordings obtained in a “normal”
pediatric population [4]. There are no adequately powered,
randomized studies to reliably confirm that the thresholds
are clinically relevant [1, 5–7].
There is a growing body of evidence highlighting the im-
portance of long-term ICP monitoring against single “snap-
shot” manometric measurements. For example, it is well-
known that standard lumbar puncture (LP) measurements
can be affected by posture, movement, emotion, anesthetic
74
agents, and hypercapnia, which can lead to unreliable estima-
tion of ICP (CSF pressure (CSFp) if it is derived from
connecting to the spinal canal rather than the parenchyma)
[8–11]. It is particularly important to bear in mind that ICP
levels are dynamic [7, 9–11] so a “normal” recorded ICP at a
single timepoint may not be a representative value. In adults
and children, the assessment of average ICP over more than
20 min (“steady state”) is reported to be more reliable than a
single opening pressure measurement [5, 10, 12–14] (Fig. 1).
Although it is important to estimate ICP more reliably, base-
line ICP as a solitary number cannot provide a full assessment
for the diagnosis and understanding of the CSF circulatory
disorders. This is because it is the production, circulation or
reabsorption of CSF that makes ICP dynamic and is likely to
be deranged in PTCS. Historically, only around 30% of these
disturbances have been shown to be reflected on steady-state
ICP values [7, 15–20]. For these reasons, monitoring of ICP in
conjunction with CSF dynamics has long been established in
the field [9, 11, 21–23].
At our center, children with suspected PTCS routinely un-
dergo a lumbar infusion study. During the infusion test, they
have a LP with measurement of a baseline, longer term CSFp
monitoring, and, if the pressure is below approximately
30 mmHg, sterile artificial CSF fluid is infused via the spinal
needle. The resultant pressure-volume data can be analyzed to
give the CSF dynamics variables such as the pulse amplitude
of ICP, the compensatory reserve, assessed mainly as the mov-
ing correlation coefficient between mean CSF pressure and
pulse amplitude (RAP), and the magnitude of slow waves of
CSFp [24–26].
Fig. 1 Variability of CSF pressure. Upper panel: pediatric patient after a
LP and connection to the computer for monitoring, showing initially
raised CSFp of 32 mmHg, spontaneously receding to 20 mmHg after
Childs Nerv Syst (2020) 36:73–86
Using this technique, children with optic disc edema at our
center can be diagnosed accurately with a raised CSFp (defi-
nite PTCS) and normal CSFp (probable PTCS more reliably
than with standard LP), and we can also characterize the CSF
dynamics of each child.
The methodology described has been used in our
center and others to report on the profiles of CSF dy-
namics that are seen in various types of pediatric and
adult hydrocephalus [5, 27–30]. By contrast, there is
little reported information on the profile of CSF dynam-
ics that characterizes PTCS. In adults with PTCS, it has
been reported that there is a dynamic coupling between
CSFp and sagittal sinus pressure (SSp), which was dem-
onstrated by performing infusion studies simultaneously
with direct cerebral venous pressure monitoring via a
catheter [31]. Our previous paper [32], which reported
on children referred to our center from 2006 to 2009,
remains as the only report on CSF dynamics in pediatric
PTCS.
We aimed to investigate the profile of CSF dynamics in
pediatric PTCS and to see if it could provide additional, clin-
ically useful insights. For this purpose, we performed a retro-
spective review of the CSF dynamics for all patients referred
to our tertiary pediatric services (≤ 16 years old) for consider-
ation of a diagnosis of PTCS from 2006 to 2016. We have re-
analyzed their original CSF infusion study data in an up-to-
date mathematical model and investigated the CSF dynamics
disturbances in definite and probable pediatric PTCS. We
have also tested their diagnostic yield compared with the tra-
ditional investigations that these children undergo.
20 min of monitoring. CSFp cerebrospinal fluid pressure, AMP
fundamental amplitude of CSFp, HR heart rate
Childs Nerv Syst (2020) 36:73–86
Methods
Patient selection
Seventy-two children were referred to the pediatric neurology
team at Cambridge University Hospital for consideration of a
diagnosis of PTCS from 2006 to 2016. Thirty-two children
from the cohort were excluded from this paper as they did not
undergo an infusion test and/or received a diagnosis other than
PTCS. Nine children were excluded (some of whom
underwent infusion) because they had PTCS due to a host of
medical conditions or medications that had previously been
reported to be associated with raised ICP (“secondary
PTCS”). In this study, we report the results of the remaining
31 children from the cohort, all of whom underwent CSFp
monitoring with or without infusion. The results of 7 of these
patients have previously been reported [32].
Patient classification
All the children who underwent CSF infusion study had
papilledema confirmed or excluded by an expert ophthalmol-
ogist and their neuroimaging reported by an expert neuroradi-
ologist. We have used here the results of that detailed analysis
using the Friedman classification (see Table 1).
Thirteen out of 31 patients were classified as definite PTCS
(group A), 14/31 as probable PTCS (group B), and /31 as not
PTCS (group C) [1]. In groups A and B, all children had
papilledema (one of the requirements) and fulfilled criteria
1–5 and 1–4 in the Friedman classification (see Table 1),
whereas none of the children in group C had papilledema.
We compared the CSF dynamics in these 3 different
groups. Because some patients did not have a fluid infusion
due to safety reasons (when baseline CSFp was too high;
above 30–40 mmHg), we studied both the baseline and the
infusion-derived parameters separately.
Infusion test
Access to the intrathecal space was gained via LP using a 18-
gauge Quincke needle, to minimize the effect of the resistance
of the needle on measured CSFp. Connection of a fluid-filled
pressure transducer (Edwards Lifesciences™) and pressure
amplifier (Spiegelberg or Philips) allowed for datapoint re-
cording at a frequency of 30–100 Hz, with the following pro-
cessing by ICM+ (University of Cambridge Enterprise Ltd).
Baseline CSF pressure was monitored for 7–15 min, followed
by infusion of Hartmann’s solution until the ICP plateaued for
10–15 min or CSFp increased to 40 mmHg or above (at which
point infusion was stopped for safety reasons). Details of the
computerized CSF infusion test have been reported in various
publications [10, 33–35].
75
Our local procedure involves attempting the infusion test in
awake children who can cooperate, with local anesthetic (li-
docaine and prilocaine cream) applied at the puncture site
45 min to 1 h before the LP. Most children inhale a 50%
nitrous oxide 50% oxygen mixture during the LP only. If a
child is not able to tolerate the above, the procedure is carried
on under GA under stable conditions (MAP, ETCO2, and
temperature) and with the use of mainly propofol and
remifentanil, with a muscle paralyzer, predominantly
rocuronium. We analyzed the results in conscious vs GA chil-
dren to identify potential differences.
All raw infusion test data was re-analyzed in a single uni-
fied model for this paper.
In all children who had a standard LP (at their local center)
followed by an infusion study (at our center), we sought the
relationship between the two values if the tests were carried
out within 3 months of each other.
ICM+ software (Cambridge enterprise) [35] was used to
calculate the CSF dynamics and compensatory reserve param-
eters: RAP which is derived from the moving correlation co-
efficient between ICP and the fundamental amplitude of ICP
(AMP) and values > 0.6 represent depleted compensatory re-
serve [36–38] and elasticity with values > 0.18 1/ml
representing depleted compensatory reserve [39, 40].
Different thresholds were also explored for these patients.
The sagittal sinus pressure (SSp) is calculated from the theo-
retical model and does not necessarily reflect the individual’s
real SSp. Thresholds for SSp and AMP have been reported as
7 mmHg and 2 mmHg, respectively [2–7]. All calculated pa-
rameters derive from known mathematical models [10, 35, 41,
42], integrated into the software. For PTCS, infusion test can-
not reliably estimate resistance to CSF outflow, as sagittal
sinus pressure usually rises with CSF pressure, making the
value of this parameter overestimated [6].
The lower breakpoint (LBP) and upper breakpoint (UBP)
of the amplitude–pressure (AMP-P) regression line was iden-
tified, when in existence, in the pediatric PTCS children and
the control groups. LBP has mainly been described in NPH, as
the breakpoint above which the AMP-P relationship starts
becoming linear. The resulting slope of the AMP-P line cor-
relates with elasticity and is a marker of depleted compensa-
tory reserve, if it is > 0.16 [43, 44]. The UBP represents a
second breakpoint, above which the same linear regression
becomes negative and possibly represents a critical point of
CSFp, above which cerebral perfusion pressure reaches the
lower limit of autoregulation; therefore, ischemia starts to de-
velop, and the pulsations transmitted from the vascular bed
decrease.
Ethical approval
Health Research Authority approval was sought and granted
and, in line with this approval and the protocol in the
 76

Table 1 Friedman classification of our 31 patients. If 5/5 criteria are fulfilled, then patients are
diagnosed with definite PTCS: (1) papilledema. (2) Normal neurological examination except for
cranial nerve abnormalities (3) neuroimaging: normal brain parenchyma on MRI for typical patients
(female and obese) and MRI ± magnetic resonance venography for others. (4) Normal CSF com-
position (5) CSFp > 25 cm H2O (20 mmHg) of 28 cmH2O if obese/anesthetized. We are showing
the demographic details and criteria of all our patients in each group. All patients fulfilled criteria 2
and 4 (not shown in the table). There was only one patient who fulfilled criteria 2–5 without
papilledema, in which case 3/4 neuroimaging characteristics shown on the right columns are
required for a diagnosis of PTCS, and they only had 2

Friedman*

Obese? Anesthetic for Pressure > 18.4 mmHg Pressure > 20.6 mmHg Sex Puberty status (Prepubertal Venography
(BMIpercentile LP/CSFIS? (≅25cmCSF)? (≅28cmCSF)? if age < 11 girls or < 12 boys) required?
>98)
Definite PTCS Y LA Y F Pubertal N
Y GA Y F Prepubertal Y
Y LA Y M Pubertal Y
N LA Y F Pubertal Y
Y LA Y F Pubertal N
Y LA Y F Pubertal N
n/a LA Y Y F Pubertal Y
N GA Y M Pubertal Y
N GA Y F Prepubertal Y
Y GA Y F Pubertal N
Y GA Y F Pubertal N
Y GA Y F Pubertal N
Y LA Y F Pubertal N
Probable PTCS Y LA N M Pubertal Y
Y LA N F Pubertal N
Y LA N F Pubertal N
Y LA N F Pubertal N
N GA N F Pubertal Y
N LA N F Pubertal Y
Y GA N M Pubertal Y
n/a GA N F Prepubertal Y
N GA N F Pubertal Y
N LA N M Prepubertal Y
Y LA N F Pubertal N
n/a LA N N F Prepubertal Y
Y LA N F Pubertal N
Not PTCS Y GA Y M Prepubertal Y
Y LA N F Pubertal N
Y LA N F Pubertal N
n/a LA N N F Pubertal Y
Y LA N F Pubertal N

Friedman*

Venography performed? Friedman Classification Empty sella Flattening of posterior Transverse venous Optic sheath dilatation
sclera sinus stenosis
Definite PTCS Y Definite PTCS
Childs Nerv Syst (2020) 36:73–86
Table 1 (continued)

Friedman*

Y Definite PTCS
Y Definite PTCS
Y Definite PTCS
Y Definite PTCS
Y Definite PTCS
Childs Nerv Syst (2020) 36:73–86

Y Definite PTCS
Y Definite PTCS
Y Definite PTCS
Y Definite PTCS
N Definite PTCS
Y Definite PTCS
Y Definite PTCS
Probable PTCS Y Probable PTCS
Y Probable PTCS
Y Probable PTCS
N Probable PTCS
Y Probable PTCS
Y Probable PTCS
N** Probable PTCS
Y Probable PTCS
Y Probable PTCS
Y Probable PTCS
N Probable PTCS
Y Probable PTCS
N Probable PTCS
Not PTCS Y Not PTCS N N Y Y
N Not PTCS
Y Not PTCS
Y Not PTCS
N Not PTCS
77
78
Cambridge University Hospitals NHS Trust, this retrospective
study was conducted without separate approval from an ethics
committee. All patients were investigated with infusion test
within the pediatric neurology clinic and neurosciences de-
partment, as a part of routine clinical assessment. They (chil-
dren/parents) all consented for these studies.
Statistical analysis
After testing for a normal distribution, non-parametric
Wilcoxon’s test was used to compare differences in CSF dy-
namics parameters between the pediatric PTCS groups. The
single-sample Wilcoxon test and t test were used to compare
CSF dynamics to their reported normative values from the
literature. The correlations between different CSF dynamics
parameters were sought using Pearson’s or Spearman’s corre-
lation coefficient, where appropriate. Multiclass ROC was
performed to test the diagnostic value of CSF dynamics vs
the 3 clinical groups using the package pROC [45]. Plots were
created, and data analysis was performed with R version 3.5.2.
Results
Demographics
Detailed demographics, of interest beyond our current focus,
such as ethnicity and BMI, are not analyzed in the current
paper. All groups were composed of a majority of female
(f:m approximately 5:1 in all 3 groups), adolescent, White,
pubertal (mean age 12 ± 3 years, similar in all 3 groups), obese
individuals. There was no obvious difference in the proportion
of severely obese patients between the probable and definite
PTCS groups. There were more prepubertal, fewer severely
obese, and more normal weight patients in the group not di-
agnosed with PTCS.
Infusion test CSFp and LPs
Three out of 13 definite PTCS children only had baseline
CSFp measurement and drainage, due to significantly raised
CSFp (around 40 or more mmHg). One patient, who did not
have papilledema or sufficient neuroimaging features to make
a diagnosis of possible PTCS, was diagnosed with chronic
migraine with raised CSFp. These children did not undergo
the infusion portion of the test for safety reasons, and as a
result, only baseline CSFp and AMP (AMPb), together with
RAP variables, could be calculated.
For the 10 children that had a CSFp measured via standard
LP prior to the infusion, the correlation between the LP CSFp
and the baseline CSFp measured was weak and non-
significant (R = 0.30; p = 0.3977, Fig. 2). All LPs had been
done 2–12 weeks prior to the infusion test.
Childs Nerv Syst (2020) 36:73–86
CSF dynamics in definite, probable, and excluded
PTCS diagnosis
The baseline features of CSF dynamics in the 3 groups are
reported analytically in Table 2.
Figure 3 shows a representative example of CSF infusion
test results in Definite PTCS vs probable and not PTCS.
CSFp was by definition increased in definite PTCS (group
A) (29.18 ± 7.72 mmHg), and significantly higher than the
probable (group B) (15.31 ± 3.47 mmHg; p = 1.644e-05) and
not PTCS (group C) (17.51 ± 5.87; p = 0.01368). The AMP
was also higher in the definite (2.18 ± 2.06 mmHg) than in
group B (0.68 ± 0.37; p = 0.01382). However, there was no
difference in either CSFp or AMP at baseline between groups
B and C (p = 0.7028 and p = 7673, respectively).
In 2 of the children with definite PTCS, the upper
breakpoint of the amplitude–pressure correlation was ob-
served at baseline (UBP at 36.36 ± 5.01 mmHg), a phenome-
non not observed in the other 2 groups. The observed pattern
of CSF dynamics in these 2 cases is demonstrated for one of
the patients in Fig. 4. Slow waves, as well as the AMP at
plateau and amp-p slope, were similar in all groups. No LBP
was observed in group A, but it was present in 2/13 and 2/5
patients in groups B and C accordingly. An UBP was present
in all groups, with no difference in its value.
There was a significantly higher SSp in definite PTCS vs
the other groups (p = 0.0014 and p = 0.007992). In pairwise
comparisons of elasticity between the groups, the only signif-
icant difference was that elasticity was significantly higher in
group A than in group C.
All parameters except for SSp did not follow a normal
distribution. In group A, SSp and elasticity were the only
ones above their reported threshold (thresholds 7 mmHg;
p = 4.2e-06 and 0.18 1/ml; p = 0.001953, respectively
[2–7]). AMPb was not higher than 2 mmHg (p = 0.6848).
In group B, only the elasticity was significantly higher than
the threshold (p = 004257). For group C, all values were
not significantly different from the published normal
range. These main CSF dynamic parameters of the 3
groups are illustrated in Fig. 5.
Relationship between the CSF dynamics parameters
in groups A–C
SSp tended to show a correlation with baseline CSFp in
groups A (R = 0.60; p value = 0.05034) and B (R = 0.59; p
value = 0.03403), but not in group C (R = −0.016; p val-
ue = 0.9833). Similarly, elasticity showed a significant cor-
relation with SSp in group A (R = 0.67; p value = 0.02807)
and group B(R = 0.59; p value = 0.03193), and not in
group C (negative correlation, not significant). There was
no correlation between CSFp and RAP or AMP with SSP
and elasticity.
Childs Nerv Syst (2020) 36:73–86 79

Fig. 2 Discordance between

CSFp derived from LP performed
either right before the CSF
infusion study or within 3 months
before the infusion study

ROC analysis between clinical classification and CSF
dynamics in pediatric PTCS
No baseline or infusion-derived parameter on its own
could give a satisfactory separation between the 3 groups.
However, given the significant differences in CSFpb, SSp,
and elasticity, and secondarily in AMPb and CSFpp, the
best AUC was when the correlation between CSFpb, SSp,
and elasticity was integrated into a linear model, with a
resulting AUC 93.8%, with 84.8–100% CI (95% CI) in
multiclass ROC analysis among the 3 different groups
(Fig. 6).

Table 2 Baseline and infusion-based CSF dynamics parameters of the 3
clinically classified pediatric PTCS groups. Values are represented as
mean ± SD. A different number of patients (N) is shown in each row
for group A, since only baseline parameters were monitored in 3/14
patients. RAP, index of compensatory reserve, from the correlation of
AMP and ICP. CSFpb, CSF pressure at baseline. AMPb, fundamental
amplitude of ICP at baseline. SSp, sagittal sinus pressure. CSFpp, CSFp
at plateau. AMPp, AMP at plateau. AMP-P slope, slope of the amplitude–
pressure line. LBP, lower breakpoint of the amplitude–pressure line. UBP,
upper breakpoint of the amplitude–pressure line

Variable A: definite PTCS B: probable p value A and B C: not PTCS (N = 5) p value A–C p value B and C
PTCS (N = 13)

Baseline CSFp (mmHg) 29.18 ± 7.72 15.31 ± 3.47 1.644e-05 17.51 ± 5.87 0.01368 0.7028
(N = 13)
AMPb (mmHg) 2.18 ± 2.06 (N = 13) 0.68 ± 0.37 0.01382 0.89 ± 1.03 0.1433 0.7673
Baseline RAP 0.58 ± 0.3 (N = 13) 0.46 ± 0.18 0.2813 0.37 ± 0.12 0.2075 0.4016
Elasticity (1/ml) 0.36 ± 0.19 (N = 10) 0.39 ± 0.26 0.9505 0.15 ± 0.06 0.002671 0.1031
SSp (mmHg) 18.99 ± 4.08 9.55 ± 11.9 0.0014 8.65 ± 1.17 0.007992 0.6928
(N = 10)
CSFpp (mmHg) 32.89 ± 2.92 25.42 ± 4.47 0.0002243 25.25 ± 6.1 0.03996 0.775
(N = 10)
AMPp (mmHg) 2.1 ± 1.11 (N = 10) 1.55 ± 0.86 0.2264 2.13 ± 1.86 0.953 0.8436
CSFpp–CSFpb (mmHg) 7.44 ± 2.73 10.11 ± 4.06 0.1661 7.74 ± 2.9 0.953 0.3873
AMPp–AMPb (mmHg) 0.93 ± 0.7 0.88 ± 0.7 0.99 1.24 ± 1.15 0.6787 0.775
Amp-p slope 0.15 ± 0.09 (N = 10) 0.09 ± 0.05 0.1621 0.13 ± 0.08 0.8539 0.3233
LBP (mmHg) NA (N = 0) 10.5 ± 2.12 (N = 2) NA 15.5 ± 3.53 (N = 2) NA 0.3333
UBP (mmHg) 36.36 ± 5.01 (N = 8) 34.5 ± 11.39 (N = 4) 0.99 29.5 ± 7.78 (N = 2) 0.5714 0.8
Slow waves at baseline (mmHg) 1.16 ± 1.43 (N = 13) 0.83 ± 0.85 0.8798 0.79 ± 0.56 0.99 0.99
Slow waves at plateau (mmHg) 3.21 ± 2.51 (N = 10) 4.9 ± 7.07 0.7501 1.5 ± 1.33 0.2828 0.5427
80
Fig. 3 Representative example of a CSF infusion study and analysis of
CSF dynamics on a pediatric patient with definite, active PTCS. a
Infusion study recording. A 10–15-min baseline is monitored to ensure
stable baseline pressure. The start of infusion is indicated with an arrow
and the infusion period is highlighted after the start on the right end. CSFp
is elevated, usually > 20 mmHg, with low resistance to CSF outflow
demonstrated from the generally low plateau of CSFp during infusion.
AMP is linearly related to CSFp, with a high correlation coefficient at
baseline (RAP > 0.4), indicating depleted compensatory reserve. RAP is
not always reliable in such short recordings, as shown by the artifacts in
AMP, corresponding to gaps in RAP calculation. b Analysis of CSF
dynamics in ICM+. Left curve: the solid, curved line represents the
theoretical model representing the response of CSFp to infusion at each
Influence of GA
Overall, CSFp monitoring with or without infusion was per-
formed under GA on 6/13 patients in group A, 4/13 in group B
and 1/5 in group C.
Childs Nerv Syst (2020) 36:73–86
time—baseline, during infusion and plateau. The dotted data points
represent the calculations performed by the interpreter of the CSF test
and which should optimally fit the theoretical mode. Right curve: The
solid line represents a similar model, where the calculations of the user
should fit the pressure-volume curve as proposed by Marmarou ([46]) and
integrated in ICM+. The dotted data points result from the user-performed
analysis of the test. CSFp CSF pressure, AMP fundamental amplitude of
CSFp. The pressure of the sagittal sinus and elasticity among others is
calculated from this mathematical model and are most reliable when the 2
curves coincide. RAP, the correlation between AMP and CSF pressure,
when available reliably at baseline, is typically elevated, showing deplet-
ed pressure-volume compensation
Within the groups, CSFpb was higher in the definite PTCS
group that received GA (34.8 ± 7.44 vs 23.06 ± 5.43; p =
0.0452); however, when the 3 children with very high baseline
pressures (who underwent drainage and no infusion study)
were removed from the analysis, the difference was no longer
Childs Nerv Syst (2020) 36:73–86
Fig. 4 Upper panel: critically high CSFp (42 mmHg), monitored in
operating theaters under general anesthesia. Note that RAP is initially
lower, representing possible ischemia cause by low cerebral perfusion
pressure (CPP). Mean arterial blood pressure was 65 mmHg; therefore,
CPP was 23 mmHg. During CSF drainage, the positive correlation
between AMP and CSFp was restored, and RAP became positive at the
higher end, representing depleted compensatory reserve. It can be
present. There were no other differences in the definite PTCS
group. In the probable PTCS group, only SSp and elasticity
appeared to be lowered by GA (0.12 ± 3.48 vs 13.16 ± 6.59;
p = 0.006851 and 0.11 ± 0.08 vs 0.53 ± 0.16; p = 0.006774.
The amplitude of slow waves was consistently suppressed
by GA at baseline and plateau within all groups.
Discussion
The diagnostic implications of measuring the full set of CSF
dynamics parameters in pediatric PTCS are highlighted in this
paper, in order to provoke further investigation and to encour-
age the use of this approach more widely in clinical practice.
CSF dynamics could provide information on the contents of
81
observed that CSFp, AMP, and HR signals after the start of drainage
contain a lot of artifacts. Lower panel: the transition CSFp point
between low CPP and restoration of normal CPP represents the upper
breakpoint of the amplitude–pressure regression line, above which the
linear correlation between AMP and CSFp is lost and tends to become
negative, as shown by the negative RAP in the upper panel
CSFp that are not obvious from CSFp as a solitary number,
accurately differentiating between the clinically and radiolog-
ically derived entities of PTCS.
We have highlighted that (A) a single recording of a raised
CSFp is not necessarily a representative value for a patient’s
ICP and (B) a raised CSFp value does not give a full under-
standing of a patient’s CSF circulation, including venous drain-
age. Despite its limitations, we acknowledge that CSFp is a
diagnostic criterion in current guidelines and we have therefore
described our precise methodology for maximizing the reliabil-
ity of this single CSF dynamics variable. This methodology
includes the measurement of “steady-state” CSFp ([8, 9, 42])
and simultaneous measurement of the blood circulation with
CSF pressure recordings, as it is known that there is a compli-
cated interaction between the CSF circulation and the cerebral
82
Fig. 5 CSF infusion study results. Summary data of CSF infusion study
results according to final diagnosis. Values are represented as mean ±
SEM. Normal thresholds for each parameter are indicated with the
horizontal black lines. a CSF pressures: CSFp at baseline and plateau
and SSp, as calculated during infusion. b AMP at baseline and plateau,
as well as compensatory reserve coefficients RAP and elasticity. The
threshold for AMPp, 4 mmHg, is not shown since it is much higher
than the average in the image and outside the scale for our calculations.
The single asterisk indicates significantly higher mean than the other
blood circulation in CSF disorders ([7, 9, 43]). We also aim to
minimize the impact of other confounding factors such as
stress, position, and sedation [1, 4, 11, 47]. These factors and
how they influence the CSFp cannot be studied and
comprehended without analysis of the CSFp components, pref-
erably with cerebral multi-modality monitoring [11, 25, 42].
Fig. 6 Area under the curve
(AUC) and 95% CI among
Friedman classification of groups
A–C and CSF dynamics, in par-
ticular the linear model of inter-
action of CSFp at baseline, elas-
ticity, and sagittal sinus pressure
Childs Nerv Syst (2020) 36:73–86
groups and from the normal threshold, double asterisks indicate
significantly higher than the other groups but not from the threshold,
and triple asterisks denote the difference in the number of patients in
group A (N = 14 vs N = 11), due to the fact that 3 patients only had
baseline values; hence, AMPb is higher than AMPp in that group.
CSFpb CSF pressure at baseline, CSFpp CSF pressure at plateau, SSp
sagittal sinus pressure, AMPb fundamental amplitude of ICP at baseline,
AMPp AMP at plateau (mmHg), RAP compensatory reserve index
Studies have already shown that there is a lack of reliability
in the manometry reading of a CSFp [8, 11], something which
is also partially shown by the lack of correlation between
manometry and baseline CSFp from a computerized infusion
test in our cohort. However, these readings were not taken at a
very close time, but up to 3 months apart. Nonetheless, they
Childs Nerv Syst (2020) 36:73–86
represent the clinical reality of trying to make a diagnosis and
a treatment plan from separated LPs and/or drainages, often
weeks and month apart from each other.
The CSFp at baseline was by definition raised > 20 mmHg
in pediatric PTCS and on average in our group > 25 mmHg. In
the group of children with definite PTCS, there were 3 children
with a CSFp > 30 mmHg, and 2 of them had critical levels of
CSFp with compromised cerebral perfusion. Although infusion
was not possible in these children, the CSF dynamics parame-
ters that could be measured were revealing; a pattern of negative
AMP-CSFp relationship was observed, indicating low-cerebral
perfusion pressure. It has been reported in traumatic brain injury
that when ICP reaches critical levels (usually > 25 mmHg in
head injury), the AMP-ICP relationship becomes negative [9,
11], which appears to also be the case with these 2 patients.
However, more data are needed in order to confirm this pattern
and association in pediatric PTCS. These cases, although seem-
ingly rare, highlight the importance of an extended multi-
parametric monitoring of the cerebral circulation, including sys-
temic arterial pressure and possibly cerebral autoregulation,
blood flow, and oxygenation. It would be interesting, for future
purposes, to adopt at a research setting initially a multi-
parameter monitoring approach, especially in suspicion of seri-
ous disease and explore the clinical implications of possible
cerebral hypoperfusion.
The difficulty in diagnosing and classifying pediatric PTCS
lies in the probable group, where the CSFp was not raised
(average 15 mmHg) and did not differ to the children in whom
PTCS was excluded. In these cases, it is of significance to
investigate the dynamics of the CSFp. Neither CSFp nor
AMP at baseline seemed to help in differentiating between
those probable and not PTCS. However, both elasticity and
SSp were significantly lower in group C and could potentially
provide a key in the differentiation between the two groups.
Both elasticity and SSp were elevated in children with definite
as well as probable PTCS, showing further potential markers of
the disease besides CSFp. However, it is important to bear in
mind that both of these parameters are subject to calculation
errors and SSp represents a value derived from mathematical
modeling of the CSF circulation and not the patient’s actual
SSp. Unfortunately, it is not possible to understand or confirm
what the probable PTCS group represents, as compared with
definite PTCS and normal children, until at least some further
follow-up and perhaps investigations are performed. Higher
SSp as well as higher elasticity could perhaps highlight a ten-
dency or the beginning of the syndrome that has not reached the
severity of the definite group.
Besides elasticity and SSp, and partially RAP, none of the
other infusion-derived parameters were above the reported
normal thresholds (0.6 for RAP—lowered to 0.4 for the this
cohort of patients, 0.16 for the amp-p line, 4 mmHg for
AMPp, 10–13 mmHg*min/ml for Rout), with the exception
of slow waves of CSFp at plateau, the magnitude of which
83
exceeded 1.5 mmHg in groups A and B, but not in group C.
Also, the absolute rise in CSFp during infusion was the same
in all groups. This rise, divided by infusion rate, denotes the
resistance to CSF outflow. As PTCS SSp follows the rise in
CSFp, this way of estimation gives an overestimated value of
resistance. Nevertheless, in PTCS, CSF circulation is probably
normal, as the gradient CSFpp–CSFpb is low [15].
For slow waves in particular, there is no well-described
threshold in any of the CSF disorders [24, 26, 48]. Although
the differences in the magnitude of slow waves between groups
were not significant, there seems to be a tendency of higher
magnitude in group A than in groups B and C, group C possibly
having the lowest magnitude. Perhaps an increased magnitude of
slow waves when the CSF circulation is challenged could reveal
a disturbed CSF circulation and assist in differentiating between
definite, probable PTCS and mimics. This will need further con-
firmation. Another finding that could provide interesting infor-
mation in excluding or confirming PTCS is the presence or ab-
sence of a LBP in the AMP-P line. None of the children in the
definite group presented with such a breakpoint, whereas it was
observed in 2 children for each of the other groups.
There was a correlation between CSFpb and SSp, but not
CSFpp. This finding could be similar to what has been shown
in adults with PTCS [31], whereby SSp and CSFp are coupled
at baseline and during infusion. This relationship, however,
cannot be validated in children without direct SSp measure-
ment and there is ethical equipoise currently between the risks
and benefits of exposing them to invasive cerebral venous
investigation. The correlation between SSp and elasticity
found mainly in our definite PTCS patients could also dem-
onstrate part of the pathophysiology of PTCS, whereby
“faulty” venous sinuses allow the transmission of the CSF
pressure onto the sinuses, leading to a coupling of the two
pressure and thereby a “stiffer” brain. It is therefore possible
in children with normal CSFp but abnormal CSF dynamics
(e.g., a depleted compensatory reserve) or an abnormal cere-
bral venous sinus system [5] that CSFp levels can rise with
minimal perturbation of the current equilibrium state.
It was intriguing to find out that no single CSF pressure or
dynamics parameter on its own could accurately differentiate
between the 3 groups with an AUC > 0.80. Our findings from
the ROC analysis show a potential overlap between diagnosis
and classification from clinical examination and neuroradiology,
with CSF dynamics parameters, in particular, CSFpb, SSP, and
elasticity. The main discrepancy between CSF dynamics and the
Friedman classification lie in the patient in group C with no
papilledema and disturbed CSF dynamics and in group B, where
there is a lot of heterogeneity among the patients. We therefore
could speculate that monitoring and infusion, not just baseline
parameters, are possibly essential for accurate patient classifica-
tion and characterization of the disturbance in the CSF circula-
tion. In this context, it is of interest to look into the case of one out
of the 5 patients without PTCS, whose CSFp was 27 mmHg and
84
did not meet the criteria for PTCS from the Friedman classifica-
tion. Therefore, in the absence of papilledema, even a confirmed
intracranial hypertension perhaps is not synonymous to the actual
syndrome and the venous compartment in such cases could merit
further investigation. Moreover, it may be possible to distinguish
between these two entities both clinico-radiologically, as well as
with monitoring of CSF dynamics, and more patients will be
needed to make these distinctions.
Finally, the influence of GA on CSF dynamics could not be
clearly demonstrated in this small cohort. It has been reported in
NPH and TBI patients that GA possibly has no effect on base-
line CSFp or elasticity, but significantly dampens the magnitude
of slow waves [25]. This study was derived from the same
center and therefore included the same GA protocol as the
NPH group (propofol + remifentanil and a muscle relaxant,
usually rocuronium, naturally with different doses in adults vs
pediatric patients). The fact that CSFpb was increased in defi-
nite PTCS under GA vs conscious children could be a random
finding, or show that anesthetic agents could increase CSFp in
pediatric PTCS patients, which is not justified from the litera-
ture on the influence of anesthetic agents such as propofol on
cerebral blood flow and metabolism. Additionally, this finding,
as well as similar previous reports [4, 25, 47], could highlight
that children needing GA are usually the ones with a higher
BMI, that in our experience is less likely to tolerate a LP, and
could reflect the known relationship with BMI in PTCS sever-
ity as well as the influence of increased abdominal pressure on
CSFp. Preliminarily derived from our set, 7/14 children needed
GA in group A, 3/3 with CSFp> 30 mmHg. In comparison, our
adult and pediatric hydrocephalus children normally require
GA when their symptoms are very prominent, which prohibits
them from complying with investigations. Similarly, the lower
SSp and elasticity in anesthetized probable PTCS patients could
mean several things, but this study is not adequately powered to
identify if this effect is real. Notably, 14/39 (36%) of our total
children (including the ones diagnosed with secondary PTCS
and were excluded from this analysis) that underwent an infu-
sion test had GA during the investigation. This percentage was
lower compared with a reported national average of 45% [2],
and we utilized our longer term average CSFp technique that
allowed for reliable CSF pressure measurement on a truly con-
tinuous scale.
Limitations Our main limitation is the small number of pa-
tients in each group and subgroup. Although we were able
to show statistically significant results with p < 0.001, sug-
gesting that the effect sizes are large, analysis of a larger co-
hort is needed to generalize these results to all children with
PTCS. Even though we tried to avoid further confusion by
excluding secondary PTCS and selecting well-matched con-
trol groups, there were still a lot of different patterns and cases
that will need further elaboration in the future, as well as more
Childs Nerv Syst (2020) 36:73–86
thoroughly designed studies on the influence of GA on CSF
dynamics.
Connected to the above is the fact that to date we have no
definitive normative data for CSF dynamics in children. We only
performed an infusion test in 5 children without PTCS—initially,
they were considered to have a disc appearance that was consid-
ered to be borderline for edema, but this was later clarified and
confirmed to not be disc edema. As such, the “abnormal” thresh-
olds for the CSF dynamics parameters have mainly been de-
scribed in hydrocephalus and TBI. From this preliminary study,
thresholds appear similar; however, more studies, adequately
powered, are needed to validate these parameters.
RAP has got a calculation window of 4 min, and in the short
time of the infusion test, the smallest artifacts could make the
calculation unreliable; however, calculation without artifacts was
possible in all 31 patients. Longer term monitoring is preferred,
and RAP is most reliable then; however, a good 10-min baseline
with stable variables could give us an estimation of RAP and the
compensatory reserve. The magnitude of slow waves is strongly
influenced by GA [25] and would require further investigation in
awake pediatric PTCS patients.
Conclusion
CSF dynamics in PTCS show a peculiar profile: baseline CSF
pressure is elevated and estimated sagittal sinus pressure is
elevated. Elasticity and compensatory reserve are depleted.
CSF circulation is probably undisturbed. Very high CSF pres-
sure may contribute to low cerebral perfusion, exposing pa-
tients to chronic sub-acute ischemia.
Compliance with ethical standards
Health Research Authority approval was sought and granted and, in line
with this approval and the protocol in the Cambridge University Hospitals
NHS Trust, this retrospective study was conducted without separate ap-
proval from an ethics committee. All patients were investigated with
infusion test within the pediatric neurology clinic and neurosciences de-
partment, as a part of routine clinical assessment. They (children/parents)
all consented for these studies.
Conflict of interest Author MC has a partial financial interest in licens-
ing ICM+ software (Cambridge Enterprise).
Open Access This article is distributed under the terms of the Creative
Commons Attribution 4.0 International License (http://
creativecommons.org/licenses/by/4.0/), which permits unrestricted use,
distribution, and reproduction in any medium, provided you give appro-
priate credit to the original author(s) and the source, provide a link to the
Creative Commons license, and indicate if changes were made.
Childs Nerv Syst (2020) 36:73–86
References
1. Friedman D, Liu GT, Digre KB (2013) Revised diagnostic criteria
for the pseudotumor cerebri syndrome in adults and children.
Neurology 81:1159–1165
2. Matthews Y-Y, Dean F, Lim MJ et al (2017) Pseudotumor cerebri
syndrome in childhood: incidence, clinical profile and risk factors
in a national prospective population-based cohort study. Arch Dis
Child 102:715 LP–715721
3. Moavero R, Sforza G, Papetti L, Battan B, Tarantino S, Vigevano F,
Valeriani M (2018) Clinical features of pediatric idiopathic intracra-
nial hypertension and applicability of new ICHD-3 criteria. Front
Neurol 9:819
4. Avery RA, Shah SS, Licht DJ, Seiden J (2010) Reference range for
cerebrospinal fluid opening pressure in children. N Engl J Med
363(9):891–893
5. Kasprowicz M, Lalou DA, Czosnyka M, Garnett M, Czosnyka Z
(2016) Intracranial pressure, its components and cerebrospinal fluid
pressure–volume compensation. Acta Neurol Scand 134:168–180
6. Pickard JD, Czosnyka Z, Czosnyka M et al (2008) Coupling of
sagittal sinus pressure and cerebrospinal fluid pressure in idiopathic
intracranial hypertension–a preliminary report. Acta Neurochir
Suppl 102:283–285
7. Pickard J (1988) Physiology and pathophysiology of the cerebro-
spinal fluid. J Neurol Neurosurg Psychiatry 51:469–470
8. Cartwright C, Igbaseimokumo U (2015) Lumbar puncture opening
pressure is not a reliable measure of intracranial pressure in chil-
dren. J Child Neurol 30(2):170–173
9. Czosnyka M, Hutchinson PJ, Balestreri M et al (2006) Monitoring
and interpretation of intracranial pressure after head injury. Acta
Neurochir Suppl 96:114–118
10. Czosnyka M, Czosnyka Z, Momjian S, Pickard JD (2004)
Cerebrospinal fluid dynamics. Physiol Meas 25:R51–R76
11. Owler BK, Fong KC, C Z (2001) Importance of ICP monitoring in
the investigation of CSF circulation disorders. Br J Neurosurg 15:
439–440
12. Czosnyka M (1996) Testing of cerebrospinal compensatory reserve
in shunted and non-shunted patients:a quide to interpretation based
on an observationl study. J Neurol Neurosurg Psychiatry 60(5):
549–558
13. Czosnyka M, Pickard JD (2004) Monitoring and interpretation of
intracranial pressure. J Neurol Neurosurg Psychiatry 75:813–821
14. Schuhmann MU, Sood S, McAllister JP et al (2008) Value of over-
night monitoring of intracranial pressure in hydrocephalic children.
Pediatr Neurosurg 44(4):269–279
15. Ekstedt J a N (1978) CSF hydrodynamic studies in man. J Neurol
Psychiatry 41:345–353
16. Børgesen SE, Albeck MJ, Gjerris F et al (1992) Computerized
infusion test compared to steady pressure constant infusion test in
measurement of resistance to CSF outflow. Acta Neurochir (Wien)
119:12–16
17. Nabbanja E, Czosnyka M, Keong NC, Garnett MR, Pickard JD,
Lalou AD (2018) Is there a link between ICP-derived infusion test
parameters and outcome after shunting in NPH ? Acta Neurochir
Suppl 126:229–232
18. Momjian S, Owler BK, Czosnyka Z, Czosnyka M, Pena A, Pickard
JD (2004) Pattern of white matter regional cerebral blood flow and
autoregulation in normal pressure hydrocephalus. Brain 127(Pt5):
965–972
19. Lalou AD, Czosnyka M, Donnelly J et al (2018) Cerebral autoreg-
ulation, cerebrospinal fluid outflow resistance, and outcome follow-
ing cerebrospinal fluid diversion in normal pressure hydrocephalus.
J Neurosurg 38(7):587–592
20. Marmarou A, Maset AL, Ward JD, Chois S, Brooks D, Lutz HA,
Moulton RJ, Muizelaar JP, DeSalles A, Young H (1987)
85
Contribution of CSF and vascular factors to elevation of ICP in
severely head-injured patients. J Neurosurg 66:883–890
21. Czosnyka M, Piechnik S, Richards HK, Kirkpatrick P, Smielewski
P, Pickard JD (1997) Contribution of mathematical modelling to the
interpretation of bedside tests of cerebrovascular autoregulation. J
Neurol Neurosurg Psychiatry 63:721–731
22. Donnelly J, Budohoski KP, Smielewski P, Czosnyka M (2016)
Regulation of the cerebral circulation : bedside assessment and
clinical implications. Crit Care 20(1):129
23. Sykora M, Czosnyka M, Liu X, et al (2016) Autonomic impairment
in severe traumatic brain injury: a multimodal neuromonitoring
study. 44(6):1173–1181
24. Lemaire JJ, Khalil T, Cervenansky F et al (2002) Slow pressure
waves in the cranial enclosure. Acta Neurochir 144:43–254
25. Lalou DA, Czosnyka M, Donnelly J, Lavinio A, Pickard JD,
Garnett M, Czosnyka Z (2016) Influence of general anaesthesia
on slow waves of intracranial pressure. Neurol Res 38:587–592
26. Czosnyka M, Laniewski W, Batorski L, Zaworski W (1988)
Analysis of intracranial pressure waveform during infusion test.
Acta Neurochir 93:140–145
27. Lofgren JAN, Zwetnow N (1973) The pressure-volume curve of the
cerebrospinal fluid space in dogs. Acta Neurol Scand 49:557–574
28. Swallow DMA, Fellner N, Varsos GV, Czosnyka M, Smielewski P,
Pickard JD, Czosnyka Z (2014) Repeatability of cerebrospinal fluid
constant rate infusion study. Acta Neurol Scand 130:131–138
29. Tisell M, Edsbagge M, Stephensen H, Czosnyka MWC (2002)
Elastance correlates with outcome after endoscopic third
ventriculostomy in adults with hydrocephalus caused by primary
aqueductal stenosis. Neurosurgery 50:70–77
30. Ekstedt J (1977) CSF hydrodynamic studies in man. 1. Method of
constant pressure CSF infusion. J Neurol Neurosurg Psychiatry 40:
105–119
31. Pickard JD, Czosnyka Z, Czosnyka M et al (2009) In: Steiger H-J
(ed) Coupling of sagittal sinus pressure and cerebrospinal fluid
pressure in idiopathic intracranial hypertension – a preliminary re-
port BT - Acta Neurochirurgica Supplements. Springer Vienna,
Vienna, pp 283–285
32. Krishnakumar, Deepa. Pickard JD, Czosnyka Z, Allen L, Parker A
(2014) Idiopathic intracranial hypertension in childhood : pitfalls in
diagnosis. 56(8):749–55
33. Chari A, Czosnyka M, Richards HK, Pickard JD, Czosnyka ZH
(2014) Hydrocephalus shunt technology: 20 years of experience
from the Cambridge Shunt Evaluation Laboratory. J Neurosurg
120:697–707
34. Czosnyka Z, Czosnyka M, Owler B et al (2005) Clinical testing of
CSF circulation in hydrocephalus. Acta Neurochir Suppl 95:247–
251
35. Smielewski P, Czosnyka M, Zabolotny W, Kirkpatrick P, Richards
H, Pickard JD (1997) A computing system for the clinical and
experimental investigation of cerebrovascular reactivity. Int J Clin
Monit Comput 14:185–198
36. Kim D, Kasprowicz M, Carrera E (2009) The monitoring of relative
changes in compartmental compliances of brain. Physiol Meas
30(7):647–659
37. Kim D-J, Eng M, Czosnyka Z et al (2009) Index of cerebrospinal
fluid compensatory reserve in hydrocephalus. Neurosurgery 64:
494–502
38. Weerakkody RA, Czosnyka M, Schuhmann MU, Schmidt E,
Keong N, Santarius T, Pickard JD, Czosnyka Z (2011) Clinical
assessment of cerebrospinal fluid dynamics in hydrocephalus.
Guide to interpretation based on observational study. Acta Neurol
Scand 124:85–98
39. Maksymowicz M, Czosnyka M, Koszewski W, Szymanska A,
Traczewski W (1989) The role of cerebrospinal compensatory pa-
rameters in the estimation of functioning of implanted shunt system
86
in patients with communicating hydrocephalus (preliminary re-
port). Acta Neurochir 101:112–116
40. Eisentrager A, Sobey I, Czosnyka M (2013) Parameter estimations
for the cerebrospinal fluid infusion test. Math Med Biol 30:157–174
41. Andersson N (2007) Cerebrospinal fluid infusion methods, online
PhD these at https://www.diva-portal.org/smash/get/diva2:140775/
FULLTEXT01.pdf
42. Andersson N, Malm J, Bäcklund T, Eklund A (2005) Assessment of
cerebrospinal fluid outflow conductance using constant-pressure
infusion - a method with real time estimation of reliability.
Physiol Meas 26:1137–1148
43. Czosnyka M, Smielewski P, Timofeev I, Lavinio A, Guazzo E,
Hutchinson P, Pickard J (2007) Intracranial pressure : more than a
number. Neurosurg Focus 22:5–11
44. Czosnyka M, Czosnyka Z, Agarwal-harding KJ, Pickard JD (2012)
Modeling of CSF dynamics: legacy of Professor Anthony
Marmarou. Acta Neurosurg Suppl 113:9–14
Childs Nerv Syst (2020) 36:73–86
45. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC,
Müller M (2011) pROC: an open-source package for R and S+ to
analyze and compare ROC curves. BMC Bioinformatics 12:77
46. Marmarou A, Shulman K (1975) Compartmental analysis of com-
pliance and outflow resistance of the cerebrospinal fluid system. J
Neurosurg 43(5):523–534
47. Friedman D (2004) Pseudotumor cerebri. Neurol Clin 22:99–131
48. Stephensen H, Andersson N, Eklund A, Malm J, Tisell M,
Wikkelsö C (2005) Objective B-wave analysis in 55 patients with
non-communicating and communicating hydrocephalus. J Neurol
Neurosurg Psychiatry 76:965–970
Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.